Use the new bch_member->seq, sb->write_time fields to detect split brain
and kick out devices when necessary.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Add new fields for split brain detection:

 - bch_member->seq, which tracks the sequence number of the last superblock
   write that happened to each member device

 - bch_sb->write_time, which tracks the time of the last superblock write,
   to allow detection of when two members have diverged but had the same
   number of superblock writes.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

nochanges means "we cannot issue writes at all"; it's possible to go
into a pseudo read-write mode where we pin dirty metadata in memory,
which is used for fsck in dry run mode and doing journal replay on a
read only mount, but we do not want to allow an actual read-write mount
in nochanges mode.

But we do always want to allow early read-write, during recovery - this
patch clarifies that.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Now that we have dynamically resizable btree paths,
check_directory_structure() can check one path - inode up to the root -
in a single transaction.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

reattach_inode() was broken w.r.t. snapshots - we'd lookup the subvolume
to look up lost+found, but if we're in an interior node snapshot that
didn't make any sense.

Instead, this adds a dirent path for creating in a specific snapshot,
skipping the subvolume; and we also make sure to create lost+found in
the root snapshot, to avoid conflicts with lost+found being created in
overlapping snapshots.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

refactoring the BTREE_ITER_WITH_UPDATES code, prep for removing the flag
and making it always-on
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

bch2_btree_iter_peek_prev() now supports BTREE_ITER_WITH_UPDATES
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

refactoring the BTREE_ITER_WITH_UPDATES code, prep for removing the flag
and making it always-on
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

XXX: we're allocating memory with btree locks held - bad

We need to plumb through an error path so we can do
allocate_dropping_locks() - but we're merging this now because it fixes
a transaction path overflow caused by indirect extent fragmentation, and
the resize path is rare.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Since the btree_paths array is now about to become growable, we have to
be careful not to refer to paths by pointer across contexts where they
may be reallocated.

This fixes the remaining btree_interior_update() paths - split and
merge.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Start to plumb through dynamically growable btree_paths; this patch
replaces most BTREE_ITER_MAX references with trans->nr_paths.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

the reflink triggers are also bumping up against the maximum number of
paths in a transaction - and generating proportional numbers of updates.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

 - Some tweaks to greatly reduce locking overhead for the list of btree
   transactions, so that it can always be enabled: leave btree_trans
   objects on the list when they're on the percpu single item freelist,
   and only check for duplicates in the same process when
   CONFIG_BCACHEFS_DEBUG is enabled

 - don't zero out the full btree_trans() unless we allocated it from
   the mempool
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Upcoming patches are going to be changing trans->paths to a
reallocatable buffer. We need to guard against use after free when it's
used by other threads; this introduces RCU protection to those paths and
changes them to check for trans->paths == NULL
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

path->idx is now a code smell: we should be using path_idx_t, since it's
stable across btree path reallocation.

This is also a bit faster, using the same loop counter vs. fetching
path->idx from each path we iterate over.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

dead code
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

reducing our usage of path->idx
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

bpos_eq() is cheaper than bpos_cmp()
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Bit of cleanup & modernization: also moving this code to util.c, it'll
be used by userspace as well.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>

dead code
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>